Rules Tracker
Verified · August 11, 2026
The platform policies, statutes, union terms, and EU obligations for real people in generative media — dated and linked to the primary source. Built for productions that want the answer before the client asks.
Every major ad platform now takes the same line: declare AI when you upload, and they scan for what you don’t.
In effect
Detection live Jun 1, 2026
Political and social-issue ads must declare photorealistic AI at buy time. Since June 2026, Meta scans every ad for third-party AI and attaches its own “AI info” label — it no longer needs a tip to catch undeclared AI.
Source · transparency.meta.comIn effect
C2PA labeling since May 2024
Ads with AI-generated or significantly AI-modified people or voices must carry the AIGC disclosure; TikTok rejects ads where it detects undisclosed AI. Uploads carrying C2PA Content Credentials are labeled automatically, and TikTok attaches credentials that persist on download.
Source · ads.tiktok.comIn effect
Expanded Apr 21, 2026
Realistic altered or synthetic content must be flagged at upload, with a dedicated checkbox and prominent disclosure for election ads. YouTube also scans for AI uses of a person’s face on behalf of creators, civic figures, and, since April 2026, entertainment talent through CAA, UTA, and WME.
Source · support.google.comFor now, federal enforcement means the FTC. The federal replica right is still a bill.
In effect
Since Oct 21, 2024
A testimonial from a person who doesn’t exist — including an AI persona presented as a real customer — violates the rule at up to $53,088 per violation. AI that changes what a real person appears to say is covered too.
Source · federalregister.govIn committee
Introduced Apr 9, 2025
The proposed federal digital-replica right: it would let people sue over unauthorized AI replicas of their voice or likeness anywhere in the country. Platforms keep a liability shield only if they remove a replica quickly after a valid notice. Cosponsors keep arriving in both chambers; committee action hasn’t.
Source · govinfo.govThirty-eight states enacted AI measures in 2025 alone (NCSL), and publicity statutes now run from Tennessee to New York. These five are the ones a production plans around.
In effect
Effective Jun 9, 2026
Any ad produced for a commercial purpose must conspicuously disclose when a synthetic performer — AI content that looks like a real human performance — appears in it. $1,000 for a first violation, $5,000 for each after. The first state disclosure law written for exactly this workflow.
Source · nysenate.govIn effect
Since Jan 1, 2025
A contract clause allowing a digital replica of a performer is unenforceable unless it specifically describes intended uses and the performer had counsel or union representation. Replicas of deceased personalities need estate consent, at $10,000 minimum liability.
Source · leginfo.legislature.ca.govIn effect
Operative Aug 2, 2026
Covered AI tools must embed an invisible disclosure in every output and offer free tools to detect it. From 2027, large platforms must surface it. Ad assets carry that built-in marker whether or not the production adds a label.
Source · leginfo.legislature.ca.govIn effect
Since Jul 1, 2024
Voice joins name and likeness as a protected property right. A recognizable AI voice clone in commercial use requires consent — and distributing tools built primarily for unauthorized cloning carries its own liability.
Source · capitol.tn.govIn effect
Since Jan 1, 2025
The Right of Publicity Act now reaches digital replicas: knowingly distributing a work with an AI recreation of a real person’s voice or likeness in a performance they never gave creates liability. That sits on top of the existing consent requirement for commercial use of identity.
Source · ilga.govIf the cast is union, the contract is the floor — and since May 2025, the floor is specific.
In effect
Ratified May 21, 2025
Consent is required before a digital replica is even created, and informed, use-specific consent before it runs. Generating a performance pays 1.5× session plus full holding and use fees. Replicas must be access-controlled and destroyed after the last commercial’s maximum period of use unless the performer agrees otherwise.
Source · sagaftra.orgFor campaigns reaching EU audiences, the transparency obligations took effect August 2.
Applicable
Since Aug 2, 2026
Deepfakes shown to EU audiences must be disclosed at first exposure, and generative tools must embed an AI marker in their outputs — standard editing and assistive tools that don’t substantially change the content are exempt. The marker means covered assets arrive pre-labeled; the disclosure is on the campaign.
Source · eur-lex.europa.euPublished
Jun 10, 2026
The Commission-endorsed way to mark and label AI-generated content, standard icons included. Following it simplifies Article 50 compliance for EU-facing campaigns.
Source · digital-strategy.ec.europa.euEnforcement doesn’t wait for statutes. The clearest recent case started inside a vendor.
Reported
Mar 24, 2026
An AI-UGC vendor’s employee forged creator licensing agreements; 33 avatars came down and every customer was notified — after real campaigns had already run on them. The advertisers had no way to know. When consent turns out to be fake, the exposure is the producer’s, not the vendor’s.
Source · rollingstone.comEvery entry links its primary source — the statute text on the legislature’s own site, the platform’s own policy page, the union’s own summary. The date above is the last time we checked every entry against its source; entries change only when their sources do. Not legal advice — before you rely on an entry, confirm it with counsel.
Tracking a rule we should have here? Tell us: hello@trypluribus.com.